DETECTION OF FRAUDULENT VEHICLE INSURANCE CLAIMS USING MACHINE LEARNING

• Main Author: Natasha, Ellen
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/54805
• Institution: Institut Teknologi Bandung

The insurance industry is inseparable from insurance fraud which causes enormous losses to insurance companies. Therefore, the detection of fraudulent insurance claims is important in order to minimize the losses caused by the fraud. In this fourth industrial revolution, machine learning can be used to detect insurance claims fraud. One of the insurance sectors that is often targeted for fraud is vehicle insurance. In this paper, fraud detection model will be made using vehicle insurance claims data. To build this model, it is necessary to determine the features that describe the characteristics of this fraud. The model built in this study can detect fraud with only 10 features. From these features, the characteristics of fraudulent vehicle insurance claims are analyzed in this paper. The purpose of this paper is to determine the best model in detecting vehicle insurance fraudulent claims. There are several methods used to detect this fraudulent claims which are logistic regression, decision tree, na¨?ve Bayes, and also ensemble of the three methods. In order to compare the performance of each method and determine the best model, this paper will use two validation methods which are AUC-ROC and confusion matrix.